Format - README.txt

README.txt

@@ -1,10 +1,11 @@
-				Primal-Dual Scene Flow for RGB-D cameras
+===========================================================================================
+                        Primal-Dual Scene Flow for RGB-D cameras
 ===========================================================================================
 This code contains an algorithm implemented on GPU to compute scene flow in real-time.
 It has been tested on Windows 7 and Ubuntu 14.04.
 
 
-				Project configuration and generation
+                             Configuration and Generation
 -------------------------------------------------------------------------------------------
 A CMakeLists.txt file is included to detect external dependencies and generate the project automatically. CUDA, OpenCV, MRPT and OpenNI2 are used, but not all are required (only CUDA is required, together with a GPU with CUDA capability of at least 2.0). 
 The project builds a "cudalib" embedding the scene flow algorithm and 2 different applications to evaluate and visualize it.
@@ -21,39 +22,39 @@ In case you don't know MRPT I encourage you to have a look at its website here:
 Detailed instructions about how to install it (or some of its modules) can be found here: http://www.mrpt.org/download-mrpt/
 
 
-				Compiling
+                                   Compiling
 -------------------------------------------------------------------------------------------
 The compiling process should be straightforward. The only problem might arise from headers that are included but the compiler cannot find. In this case you should find those files on your computer and include them with the correct path (for your machine).
 
 
-				Usage
+                                     Usage
 -------------------------------------------------------------------------------------------
 Both apps can read command line arguments. By adding the argument "--help" you will obtain a list of the accepted input arguments:
 
 --help: To show the list of arguments...
 
 --cam_mode cm: To open the RGB-D camera with VGA (cm = 1) or QVGA (cm = 2) resolution.
-				Only for "Scene-Flow-Visualization"
+               Only for "Scene-Flow-Visualization"
 
 --fps f: The desired scene flow frame rate (Hz), it  might be not achievable!! 
-				Only for "Scene-Flow-Visualization"
+         Only for "Scene-Flow-Visualization"
 
 --rows r: Number of rows at the finest level of the pyramid
-			Options: r=15, r=30, r=60, r=120, r=240, r=480 (if VGA)
+          Options: r=15, r=30, r=60, r=120, r=240, r=480 (if VGA)
 
 
 The coarsest level of the pyramid is always 15 x 20, so the total number of coarse-to-fine levels will depend on the parameter "rows". By default, rows = 240.
 
 The images that "Scene-Flow-Impair" reads must be named as "I1" (first intensity image), "I2" (second intensity image", "Z1" (first depth image) and "Z2" (second depth image) and must be located in the same directory as the executable. Furthermore, they must be saved with the following format:
 
 intensity images - 8 bit in PNG. Resolution of VGA or QVGA
-				Clue: Use cv::Mat image_name(height, width, CV_8U) and
-				cv::imwrite(filename, image_name) to store them
+                   Clue: Use cv::Mat image_name(height, width, CV_8U) and
+                         cv::imwrite(filename, image_name) to store them
 
 depth images - 16 bit monochrome in PNG, scaled by 5000. Resolution of VGA or QVGA
-				Clue: Use cv::Mat image_name(height, width, CV_16U) and
-				cv::imwrite(filename, image_name) to store them.
-				Multiply the real depth by 5000.
+               Clue: Use cv::Mat image_name(height, width, CV_16U) and
+               cv::imwrite(filename, image_name) to store them.
+               Multiply the real depth by 5000.
 
 
 The algorithm convergence is set to a fixed number of iterations at each level, which increases with the level resolution. You can change it at your convenience by modifying the variable num_max_iter[].